KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

EXPERIMENTAL RESEARCH

Unit Structure

1.0 Objectives

1.1 Introduction

1.2 Experimental Designs

a. Pre-experimental design
b. Quasi-experimental design
c. True experimental design

1.3 Factorial Design

1.4 Nested Design

1.5 Single Subject Design

1.6 Internal and External Experimental Validity

1.7 Controlling Extraneous and Intervening Variables

1|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.1 INTRODUCTION :

The experimental method in educational research is the application and adaptation

of the classical method of experimentation. It is a scientifically sophisticated method. It
provides a method of investigation to derive basic relationships among phenomena under
controlled condition or, more simply, to
identify the conditions underlying the occurrence of a given phenomenon. Experimental
research is the description and analysis of what will be, or what will occur, under
carefully controlled conditions.

Experimenters manipulate certain stimuli, treatments, or environmental conditions

and observe how the condition or behaviour of the subject is affected or changed. Such
manipulations are deliberate and systematic. The researchers must be aware of other
factors that could influence the outcome and remove or control them in such a way that it
will establish a logical association between manipulated factors and observed factors.

Experimental research provides a method of hypothesis testing. Hypothesis is the

heart of experimental research. After the experimenter defines a problem he has to
propose a tentative answer to the problem or hypothesis. Further, he has to test the
hypothesis and confirm or disconfirm it.

Although, the experimental method has greatest utility in the laboratory, it has
been effectively applied non-laboratory settings such as the classroom. The immediate
purpose of experimentation is to predict events in the experimental setting. The ultimate
purpose is to generalize the variable relationships so that they may be applied outside the
laboratory to a wider population of interest.

Characteristics of Experimental Method

There are four essential characteristics of experimental research:

(i) Control, (ii) Manipulation, (iii) Observation, and (iv) Replication.

Control : Variables that are not of direct interest to the researcher, called extraneous
variables, need to be controlled. Control refers to removing or minimizing the influence

2|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

of such variables by several methods such as: randomization or random assignment of

subjects to groups; matching subjects on extraneous variable(s) and then assigning
subjects randomly to groups; making groups that are as homogenous as possible on
extraneous variable(s); application of statistical technique of analysis of covariance
(ANCOVA); balancing means and standard deviations of the groups.

Manipulation : Manipulation refers to a deliberate operation of the conditions by the

researcher. In this process, a pre-determined set of conditions, called independent variable
or experimental variable. It is also called treatment variable. Such variables are imposed
on the subjects of experiment. In specific terms manipulation refers to deliberate
operation of independent variable on the subjects of experimental group by the researcher
to observe its effect. Sex, socio-economic status, intelligence, method of teaching,
training or qualification of teacher, and classroom environment are the major independent
variables in educational research. If the researcher, for example, wants to study the effect
of X method of teaching on the achievement of students in mathematics, the
independent variable here is the method of teaching. The researcher in this experiment
needs to manipulate X i.e. the method of teaching. In other words, the researcher has to
teach the experimental groups using X method and see its effect on achievement.

Observation : In experimental research, the experimenter observes the effect of the

manipulation of the independent variable on dependent variable. The dependent variable,
for example, may be performance or achievement in a task.

Replication : Replication is a matter of conducting a number of sub-experiments, instead

of one experiment only, within the framework of the same experimental design. The
researcher may make a multiple comparison of a number of cases of the control group
and a number of cases of the experimental group. In some experimental situations, a
number of control and experimental groups, each consisting of equivalent subjects, are
combined within a single experiment.

3|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.2 EXPERIMENTAL DESIGNS :

Experimental design is the blueprint of the procedures that enable the researcher to
test hypotheses by reaching valid conclusions about relationships between independent
and dependent variables (Best, 1982, p.68). Thus, it provides the researcher an
opportunity for the comparison as required in the hypotheses of the experiment and
enables him to make a meaningful interpretation of the results of the study. The designs
deal with practical problems associated with the experimentation such as: (i) how
subjects are to be selected for experimental and control groups, (ii) the ways through
which variables are to be manipulated and controlled, (iii) the ways in which extraneous
variables are to be controlled, how observations are to be made, and (iv) the type of
statistical analysis to be employed.

Variables are the conditions or characteristics that the experimenter manipulates,

controls, or observes. The independent variables are the conditions or characteristics that
the experimenter manipulates or controls in his or her attempt to study their relationships
to the observed phenomena. The dependent variables are the conditions or characteristics
that appear or disappear or change as the experimenter introduces, removes or changes
the independent variable. In educational research teaching method is an example of
independent variable and the achievement of the students is an example of dependent
variable. There are some confounding variables that might influence the dependent
variable. Confounding variables are of two types; intervening and extraneous variables.
Intervening variables are those variables that cannot be controlled or measured but may
influence the dependent variable. Extraneous variables are not manipulated by the
researcher but influence the dependent variable. It is impossible to eliminate all
extraneous variables, but sound experimental design enables the researcher to more or
less neutralize their influence on dependent variables.

There are various types of experimental designs. The selection of a particular

design depends upon factors like nature and purpose of experiment, the type of variables
to be manipulated, the nature of the data, the facilities available for carrying out the
experiment and the competence of the experimenter. The following categories of
experimental research designs are popular in educational research:

(i) Pre-experimental designs They are least effective and provide little or no
control of extraneous variables.
(ii) True experimental designs employ randomization to control the effects of
variables such as history, maturation, testing, statistical regression, and mortality.
(iii) Quasi-experimental designs provide less satisfactory degree of control
and are used only when randomization is not feasible.
(iv) Factorial designs- more than one independent variables can be manipulated
4|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

simultaneously. Both independent and interaction effects of two or more than two
factors can be studied with the help of this factorial design.

Symbols used :
In discussing experimental designs a few symbols are used. E Experimental
group
C Control group
X Independent variable Y Dependent
variable
R Random assignment of subjects to groups
Yb Dependent variable measures taken before experiment / treatment (pre-test)
Ya Dependent variable measures taken after experiment/ treatment (Post-test)
Mr Matching subjects and then random assignment to groups.

a. Pre-Experimental design :

There are two types of pre-experimental designs:

1. The one group pre-test post-test design :

This is a simple experimental research design without involvement of a control

group. In this design the experimenter takes dependent variable measures (Y b) before the
independent variable
(X) is manipulated and again takes its measures (Y a) afterwards: The difference if any,
between the two measurements (Yb and Y a) is computed and is ascribed to the
manipulation of X.

Post-
Pre-test Independent variable test
Yb X Ya

5|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

The experimenter, in order to evaluate the effectiveness of computer-based

instruction (CBI) in teaching of science to grade V students, administers an achievement
test to the whole class (Yb) before teaching through CBI. The test is administered over
the same class again to measure Ya. The means of Yb and Ya are compared and the
difference if any is ascribed to effect of X, i.e. teaching through CBI.
The design has the inherent limitation of using one group only. The design also
lacks scope of controlling extraneous variables like history, maturation, pre-test
sensitization, and statistical regression etc.

2. The two groups static design :

This design provides some improvement over the previous by adding a control
group which is not exposed to the experimental treatment. The experimenter may take
two sections of grade-V of one school or grade-V of one school or grade-V students of
two different schools (intact classes) as experimental and control groups respectively and
assume the two groups to be equivalent. No pre-test is taken to ascertain it.
Group Independent Variable Post-test
E X Ya
C - Ya

This design compares the post-test scores of experimental group (Y a E) that has
received experimental treatment (X) with that of control group (Y a C) that has not
received X.

The major limitation of the design is that there is no provision for establishing the
equivalence of the experimental (E) and control
(C) groups. However, since no pretest is used, this design controls for the effects of
extraneous variables such history, maturation, and pre-testing.

b. Quasi-Experimental Design :

Researchers commonly try to establish equivalence between the experimental and

control groups, the extent they are successful in doing so; to this extent the design is
valid. Sometimes it is extremely difficult or impossible to equate groups by random
selection or random assignment, or by matching. In such situations, the researcher uses
quasi-experimental design.

6|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

The Non-Equivalent Groups Design is probably the most frequently used design
in social research. It is structured like a pretest-posttest randomized experiment, but it
lacks the key feature of the randomized designs -- random assignment. In the Non-
Equivalent Groups Design, we most often use intact groups that we think are similar as
the treatment and control groups. In education, we might pick two comparable
classrooms or schools. In community-based research, we might use two similar
communities. We try to select groups that are as similar as possible so we can fairly
compare the treated one with the comparison one. But we can never be sure the groups
are comparable. Or, put another way, it's unlikely that the two groups would be as similar
as they would if we assigned them through a random lottery. Because it's often likely that
the groups are not equivalent, this designed was named the nonequivalent group design to
remind us.

Pre-test Independent Variable Post-test

Yb X Ya (Experimental)
Yb - Ya (Control)
So, what does the term "nonequivalent" mean? In one sense, it just means that
assignment to group was not random. In other words, the researcher did not control the
assignment to groups through the mechanism of random assignment. As a result, the
groups may be different prior to the study. This design is especially susceptible to the
internal validity threat of selection. Any prior differences between the groups may affect
the outcome of the study. Under the worst circumstances, this can lead us to conclude that
our program didn't make a difference when in fact it did, or that it did make a difference
when in fact it didn't.

The counterbalanced design may be used when the random assignment of

subject to experimental group and control group is not possible. This design is also
known as rotation group design. In counterbalanced design each group of subject is
assigned to experimental treatment at different times during the experiment. This design
overcomes the weakness of non-equivalent design. When intact groups are used, rotation
of groups provides an opportunity to eliminate any differences that might exist between
the groups. Since all the groups are exposed to all the treatments, the results obtained
cannot be attributed to the preexisting differences in the subjects. The limitation of this
design is that there is carry-over effect of the groups from one treatment to the next.
Therefore, this design should be used only when the experimental treatments are such
that the administration of one treatment on a group will have no effect on the next
treatment. There is possibility of boring students with repeated testing.

c. True experimental design :

True experimental designs are used in educational research because they ascertain
7|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

equivalence of experimental and control groups by random assignment of subjects to

these groups, and thus, control the effects of extraneous variables like history, maturation,
testing, measuring instruments, statistical regression and mortality. This design, in
contrast to pre-experimental design, is a better and used in educational research wherever
possible.

1. Two groups, randomized subjects, post-test only design:

This is one of the most effective designs in minimizing the threats to experimental
validity. In this design subjects are assigned to experimental and control groups by
random assignment which controls all possible extraneous variables, e.g. testing,
statistical regression, mortality etc. At the end of experiment the difference between the
mean post-test scores of the experimental and control group are put to statistical test t
test or analysis of variance (ANOVA). If the differences between the means are found
significant, it can be attributed to the effect of (X), the independent variable.

Group Independent Variable Post-test

E X Ya
R
C - Ya

The main advantage of this design is random assignment of subjects to groups,

which assures the equivalence of the groups prior to experiment. Further, this design, in
the absence of pretest, controls the effects of history, maturation and pre-testing etc.

This design is useful in the experimental studies at the pre-primary or primary

stage and the situations in which a pre-test is not appropriate or not available.

8|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

2. Two groups, randomized matched subject, post-test only design :

This design, instead of using random assignment of subjects to experimental and

control group, uses the technique of matching. In this technique, the subjects are paired
so that their scores on matching variable(s), i.e. the extraneous variable(s) the
experimenter wants to control, are as close as possible. One subject of each pair is
randomly assigned to one group and the other to the second group. The groups are
designated as experimental and control by random assignment (tossing a coin).

Group Independent Variable Post-test

E X Ya
MR C - Ya
This design is mainly used where two groups randomized subjects, post-test
only design is not applicable and where small groups are to be used. The random
assignment of subjects to groups after matching adds to the strength of this design. The
major limitation of the design is that it is very difficult to use matching as a method of
controlling extraneous variables because in some situations it is not possible to locate a
match and some subjects are excluded from the experiment.

3. Two groups randomized subjects, pre-test post-test design :

In this design subjects are assigned to the experimental group and the control
group at random and are given a pre-test (Y b). The treatment is introduced only to the
experimental group, after which the two groups are measured on dependent variable.
The difference in scores or gain scores (D) in respect of pre-test and post-test (Y a Yb
= D) is found for each group and the difference in scores of both the groups (D e and Dc)
is compared in order to ascertain whether the experimental treatment produced a
significant change. Unless the effect of the experimental manipulation is strong, the
analysis of the differential score is not advisable (Kerlinger, 1973, p-336). If they are
analyzed, however, a t or F test is used.

Group Pre-test Independent Variable Post-test

E Yb X Ya
C Yb - Ya

9|Page
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

The main advantages of this design include:

Through initial randomization and pre-testing equivalence between the two groups
can be ensured.
Randomization seems to control most of the extraneous variables.

But the design does not guarantee external validity of the experiment as the pretest
may increase the subjects sensitivity to the manipulation of X.

4. The Solomon three groups design :

This design, developed by Solomon seeks to overcome the difficulty of the design:
Randomized Groups, Pre-test Posttest Design, i.e. the interactive effects of pre-testing
and the experimental manipulation. This is achieved by employing a second control
group (C2) which is not pre-tested but is exposed to the experimental treatment (X).

Group Pre-test Independent Variable Post-test

E Yb X Ya
C1 Yb - Ya
C2 - X Ya
This design provides scope for comparing post-tests (Ya) scores for the three
groups. Even though the experimental group has a significantly higher mean score as
compared to that of the first control group (Y aE > Ya C1), one cannot be confident that
this difference is due to the experimental treatment (X). It might have occurred because
of the subjects pre-test sensitization. But, if the mean Y a scores of the second control
group is also higher as compared to that of the first control group (Y a C 2 > Ya C l), then
one can assume that the experimental treatment has produced the difference rather than
the pre-test sensitization, since C2 is not pre-tested.

5. The Solomon four group design :

This design is an extension of Solomon three group design and is really a

combination of two two-groups designs: (i) Two groups randomized subjects pre-test
post-test design; and (ii) Two group randomized subjects post-test only design. This
design provides rigorous control over extraneous variables and also provides opportunity
for multiple comparisons to determine the effects of the experimental treatment (X).
In this design the subjects are randomly assigned to the four groups. One
experimental (E) and three control (Cl, C 2 and C3). The experimental and the first control
group (E and C1) are pre-tested groups, and the second and third control groups (C 2 and
10 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

C3) are not pre-tested groups. If the post-test mean scores of experimental group (Y a E) is
significantly greater than the post-test mean score of the first control group (Y a C1); and
also the post test mean score of the second control group (Y a C 2) is significantly greater
than the post-test mean score of the third control group (Y a C3), the experimenter arrives
at the conclusion that the experimental treatment (X) has effect.

Group Pre- Independent Post test

test Variable
E Yb X Ya
C1 Yb - Ya
R
C2 - X Ya
C3 - - Ya
This design is considered to be strong one as it actually involves conducting the
experiment twice, once with pre-test and once without pre-test. Therefore, if the results of
these two experiments are in agreement, the experimenter can have much greater
confidence in his findings. The design seems to have two sources of weakness. One is
practicability as it is difficult to conduct two simultaneous experiments, and the
researcher encounters the difficulty of locating more subjects of the same kind. The other
difficulty is statistical. Since the design involves four sets of measures for four groups
and the experimenter has to make comparison between the experimental and first control
group and between second and third control groups there is no single statistical procedure
that would make use of the six available measures simultaneously.

11 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.3 FACTORIAL DESIGN :

Experiments may be designed to study simultaneously the effects of two or more

variables. Such an experiment is called factorial experiment. Experiments in which the
treatments are combinations of levels of two or more factors are said to be factorial. If all
possible treatment combinations are studied, the experiment is said to be a complete
factorial experiment. When two independent factors have two levels each, we call it as
2x2 (spoken "two-by-two) factorial design. When three independent factors have two
levels each, we call it 2x2x2 factorial design. Similarly, we may have 2x3, 3x3, 3x4,
3x3x3, 2x2x2x2, etc.

Simple Factorial Design :

A simple factorial design is 2x2 factorial design. In this design there are two
independent variables and each of the variables has two levels. One advantage is that
information is obtained about the interaction of factors. Both independent and interaction
effects of two or more than two factors can be studied with the help of this factorial
design.

In factorial designs, a factor is a major independent variable. In this example we

have two factors: methods of teaching and intelligence level of the students. A level is a
subdivision of a factor. In this example, method of teaching has two levels and
intelligence has two levels. Sometimes we depict a factorial design with a numbering
notation. In this example, we can say that we have a 2 x 2 (spoken "two-by-two)
factorial design. In this notation, the number of numbers tells us how many factors there
are and the number values tell how many levels. The number of different treatment
groups that we have in any factorial design can easily be determined by multiplying
through the number notation. For instance, in our example we have 2x2 = 4 groups. In
our notational example, we would need 3 x 4 = 12 groups. Full factorial experiment is an
experiment whose design consists of two or more factors, each with discrete possible
values or "levels", and whose experimental units take on all possible combinations of
these levels across all such factors. A full factorial design may also be called a fully-
crossed design. Such an experiment allows studying the effect of each factor on the
response variable, as well as the effects of interactions between factors on the response
variable.

For the vast majority of factorial experiments, each factor has only two levels.
For example, with two factors each taking two levels, a factorial experiment would
have four treatment combinations in total, and is usually called a 22 factorial design.
The first independent variable, which is manipulated, has two values called the
experimental variable. The second independent variable, which is divided into levels,
may be called control variable. For example, there are two experimental treatments,

12 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

that is, teaching through co-operative learning and teaching through lecture method. It
is observed that there may be differential effects of these methods on different levels of
intelligence of the students. On the basis of the IQ score the experimenter divides the
students into two groups: one high intelligent group and the other the low intelligent
group. There are four groups of students within each of the two levels of intelligence.

High Intelligence Low Intelligence

Group Group
Teaching Through Gain Score on the Gain Score on the
Co-operative Dependent Variable Dependent Variable
Learning Method
Teaching Through Gain Score on the Gain Score on the
Lecture Method Dependent Variable Dependent Variable

Since one of the objectives is to compare various combinations of these groups,

the experimenter has to obtain the mean scores for each row and each column. The
experimenter can first study the main effect of the two independent variables and the
interaction effect between the intelligence level and teaching method.

13 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.4 NESTED DESIGN :

In a nested design, each subject receives one, and only one, treatment condition. In a
nested design, the levels of one factor appear only within one level of another factor. The
levels of the first factor are said to be nested within the level(s) of the second factor.
When variables such as race, income and education, etc. may be found only at a
particular level of the independent variable, these variables are called nested variables. In
these studies the various nested variables are grouped for the study. For example, a
researcher is studying school effectiveness with academic achievement of students as the
indicator or criterion variable. In this type of research, school type can be nested within
individual schools which can be nested within classrooms. The major distinguishing
feature of nested designs is that each subject has a single score. The effect, if any, occurs
between groups of subjects and thus the name Between Subjects is given to these
designs. The relative advantages and disadvantages of nested designs are opposite those
of crossed designs. First, carry over effects are not a problem, as individuals are measured
only once. Second, the number of subjects needed to discover effects is greater than with
crossed designs. Some treatments by their nature are nested. The effect of gender, for
example, is necessarily nested. One is either a male or a female, but not both. Religion is
another example. Treatment conditions which rely on a pre-existing condition are
sometimes called demographic or blocking factors.

Crossed Design :

In a crossed design each subject sees each level of the treatment conditions. In a
very simple experiment, such as one that studies the effects of caffeine on alertness, each
subject would be exposed to both a caffeine condition and a no caffeine condition. For
example, using the members of a statistics class as subjects, the experiment might be
conducted as follows. On the first day of the experiment, the class is divided in half with
one half of the class getting coffee with caffeine and the other half getting coffee without
caffeine. A measure of alertness is taken for each individual, such as the number of yawns
during the class period. On the second day the conditions are reversed; that is, the
individuals who received coffee with caffeine are now given coffee without and vice-
versa. The size of the effect will be the difference of alertness on the days with and
without caffeine.

14 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

The distinguishing feature of crossed designs is that each individual will have
more than one score. The effect occurs within each subject, thus these designs are
sometimes referred to as within subjects designs.

Crossed designs have two advantages. One, they generally require fewer subjects,
because each subject is used a number of times in the experiment. Two, they are more
likely to result in a significant effect, given the effects are real.

Crossed designs also have disadvantages. One, the experimenter must be

concerned about carry-over effects. For example, individuals not used to caffeine may
still feel the effects of caffeine on the second day, even though they did not receive the
drug. Two, the first measurements taken may influence the second. For example, if the
measurement of interest was score on a statistics test, taking the test once may influence
performance the second time the test is taken. Three, the assumptions necessary when
more than two treatment levels are employed in a crossed design may be restrictive.

15 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.5 SINGLE FACTOR EXPERIMENT :

Many experiments involve single treatment or variable with two or more levels.
First, a group of experimental subjects may be divided into independent groups, using a
random method. Different treatment may be applied to each group. One group may be a
control group, a group to which no treatment is applied. For meaningful interpretation of
experiment, results obtained under treatment may be compared with results obtained in
the absence of treatment. Comparison may be made between treatments and between
treatment and a control.

Some single factor experiments involve a single group of subjects. Each subject
receives treatments. Repeated observations or measurements are made on the same
subjects.

Some single factor experiments may consists of groups that are matched on one or
more variables which are known to be correlated with the dependent variable. For
example IQ may be correlated with achievement.

An example of single factor Experiment :

It is believed that the amount of time a player warms up at the beginning will have
a significant impact on his game, lawn tennis. The hypothesis is that if he does not warm
up at all or only for a brief time (less than 15 minutes), he will be stiff and his score will
be poor. However, if he warms up too much (over 40 minutes), he will be tired and his
game score will also suffer. He needs to choose levels of warming up to test this
hypothesis that are significantly different enough. The levels he will test are warming up
for 0, 15, 30, and 45 minutes.

16 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.6 INTERNAL AND EXTERNAL EXPERIMENTAL VALIDITY :

Validity of experimentation :

An experiment must have two types of validity: internal validity and external
validity (Campbell and Stanley, 1963):

Internal validity :

Internal validity refers to the extent to which the manipulated or independent

variables actually have a genuine effect on the observed results or dependent variable and
the observed results were not affected by the extraneous variables. This validity is
affected by the lack of control of extraneous variables.
External validity :

External validity is the extent to which the relationships among the variables can
be generalized outside the experimental setting like other population, other variables.
This validity is concerned with the generalizability or representativeness of the findings
of experiment, i.e. to what population, setting and variables can the results of the
experiment be generalized.

Factors affecting validity of experimentation :

In educational experiments, a number of extraneous variables influence the results

of the experiment in way that are difficult to evaluate. Although these extraneous
variables cannot be completely eliminated, many of them can be identified. Campbell and
Stanley (1963) have pointed out the following major variables which affect significantly
the validity of an experiment:

History : The variables, other than the independent variables, that may occur between the
first and the second measurement of the subjects (Pre-test and post test).

Maturation : The changes that occur in the subjects over a period of time and confused
with the effects of the independent variables.

Testing : Pre-testing, at the beginning of an experiment, may be sensitive to subjects,

which may produce a change among them and may affect their post-test performance.

Measuring Instruments : Different measuring instruments, scorers, interviewers or the

observers used at the pre and post testing stages; and unreliable measuring instruments or
techniques are threats to the validity of an experiment.

17 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

Statistical regression : It refers to the tendency for extreme scores to regress or move
towards the common mean on subsequent measures. The subjects who scored high on a
pre-test are likely to score relatively low on the retest whereas the subjects who scored
low on the pre-test are likely to score high on the retest.

Experimental mortality : It refers to the differential loss of subjects from the

comparison groups. Such loss of subjects may affect the findings of the study. For
example, if some subjects in the experimental group who received the low scores on the
pre-test drop out after taking the test, this group may show higher mean on the post-test
than the control group.

Differential selection of subjects : It refers to difference between/among groups on

some important variables related to the dependent variable before application of the
experimental treatment.

18 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

1.7 CONTROLLING EXTRANEOUS AND INTERVENING

VARIABLES :

All experimental designs have one central characteristic: they are based on
manipulating the independent variable and measuring the effect on the dependent
variable. Experimental designs result in inferences drawn from the data that explain the
relationships between the variables.

The classic experimental design consists of the experimental group and the control
group. In the experimental group the independent variable is manipulated. In the control
the dependent variable is measured when no alteration has been made on the independent
variable. The dependent variable is measured in the experimental group the same way,
and at the same time, as in the control group.

The prediction is that the dependent variable in the experimental group will
change in a specific way and that the dependent variable in the control group will not
change.

Controlling Unwanted Influences :

To obtain a reliable answer to the research question, the design should eliminate
unwanted influences. The amount of control that the researcher has over the variables
being studied varies, from very little in exploratory studies to a great deal in experimental
design, but the limitations on control must be addressed in any research proposal.

These unwanted influences stem from one or more of the following: extraneous
variables, bias, the Hawthorne effect, and the passage of time.

Extraneous Variables :

Extraneous variables are variables that can interfere with the action of the
independent variable. Since they are not part of the study, their influence must be
controlled.

In the research literature, the extraneous variables also referred to as intervening

variables, directly affect the action of the independent variable on the dependent
variables. Intervening variables are those variables that occur in the study setting. They
include economic, physical, and psychological variables. Therefore, it is important to
control extraneous variables to study the effect of independent variable on dependent
variable. We must be very careful to control all possible extraneous variables that might
intervene the dependant variable.

19 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

Methods of controlling extraneous variables include :

randomization
homogeneous sampling techniques
matching
building the variables into the design
statistical control

Randomization : Theoretically, randomization is the only method of controlling all

possible extraneous variables. The random assignment of subjects to the various
treatment and control groups means that the groups can be considered statistically equal
in all ways at the beginning of the experiment. It does not mean that they actually are
equal for all variables.

However, the probability of their being equal is greater than the probability of their
not being equal, if the random assignment was carried out properly. The exception lies
with small groups where random assignment could result in unequal distribution of
crucial variables. If this possibility exists, the other method would be more appropriate.
In most instances, however, randomization is the best method of controlling extraneous
variables.

A random sampling technique results in a normal distribution of extraneous

variables in the sample; this approximates the distribution of those variables in the
population. The purpose of randomization is to ensure a representative sample.
Randomization comes into play when we randomly assign subjects to
experimental and control groups, thus ensuring that the groups are as equivalent as
possible prior to the manipulation of the independent variable. Random assignment
assures that the researcher is unbiased. Instead, assignment is predetermined for each
subject.

Homogeneous Sample : One simple and effective way of controlling an extraneous

variable is not to allow it to vary. We may choose a sample that is homogenous for that
variable. For example, if a researcher believes that gender of the subject might affect the
dependant variable, he/she could select the subjects of the desired gender only. If the
researcher believes that socio-economic status might influence the dependant variable,
he/she would select subject from a particular range of socio-economic status. After
selecting students from a homogenous population the researcher may assign the subjects
to experimental and control group randomly.

Matching : When randomization is not possible, or when the experimental groups are too

20 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

small and contain some crucial variables, subjects can be matched for those variables.
The experimenter chooses subjects who match each other for the specified variables. One
of these matched subjects is assigned to the control group and the other to the
experimental group, thus ensuring the equality of the groups at the outset.

The process of matching is time consuming and introduces considerable

subjectivity into sample selection. Therefore, it should be avoided whenever possible. If
we use matching, limit the number of groups to be matched and keep the number of
variables for which the subjects are matched low. Matching with more than five variables
becomes extremely cumbersome, and it is almost impossible to find enough matched
partners for the sample. Matching may be used in all research designs when we are
looking at certain outcomes and want to have as much control as possible.

Building Extraneous Variables into the Design : When extraneous variables cannot be
adequately controlled by randomization, they can be built into the design as independent
variables. They would have to be added to the purpose of study and tested for
significance along with other variables. In this way, their effect can be measured and
separated from the effect of the independent variable.

Statistical Control : In experimental designs, the effect of the extraneous variables can
be subtracted statistically from the total action of the variables. The technique of analysis
of covariance (ANCOVA) may be used for this purpose. Here, one or more extraneous
variables are measured along with the dependant variables. This method adds to the cost
of the study because of the additional data collection and analysis required. Therefore, it
should be used only as a last resort.

21 | P a g e
 KSP 6063 - RESEARCH METHODS IN SCIENCE EDUCATION

References :

Anderson, G (1990): Fundamentals of Educational Research: The Falmer Press, London.

th
Best, J.W. & Kahn, J.V. (1993): Research in Education; 7 Ed. Prentice Hall of India
Pvt., Ltd., New Delhi.

Campbell, D. T. & Stanley, J. C. (1963). Experimental and Quasi-Experimental Designs

for Research.. Chicago: Rand McNally.

Fisher, R. A. (1959). Statistical Methods & Scientific Inference. New York: Hafner
Publishing.

Gay, L.R. (1987). Educational Research, Englewood Cliffs NJ: Macmillan Publishing
Company.

nd
Kerlinger, F.N. (1964) : Foundations of Behavioural Research (2 Ed.), Surjeet
Publications, New Delhi.

nd
Koul, L. (1984): Methodology of Educational Research (2 Ed.), Vikash Publishing
House Pvt. Ltd., New Delhi.

Stockburger, David W. (1998): Introductory Statistics: Concepts, Models, and

nd
Applications, (2 Ed), www.atomicdogpublishing.com

22 | P a g e